MUSINGS / MEASUREMENTS: Multiple subwoofers to reduce nulls. The USB/UFO faithful. And thanks for the blind test submissions!

A few weeks back, I ran into this interesting article - "Is the room the most important component?".

In summary, it seems like the article is basically saying that we can rearrange our gear or perhaps treat our rooms in ways that sound good, thus making the room less important than a general consensus might suggest. There seems to be an undertone of trying to downplay the role of the sound room and suggesting that one can spend more money on high quality gear and still benefit. OK, sure, to some extent that's true; but there are obviously limits. After all, if the room is way too small, cubical, highly reflective and impractical to treat, speakers inappropriately shoved deeply against walls, or if there is no space behind the seating position, there's obviously no point spending $$$$ on gear that one can barely appreciate! In my opinion, the quality and size of the room and quality of the gear (especially speakers) should be reasonably balanced.

IMO, without doubt, the room does play a major role in the sound quality; I think it would be silly to suggest otherwise even though I have heard some completely deny this over the years! For example, objectively we can easily show the nodes (nulls) and antinodes (peaks) resulting from reflections and standing waves in our small domestic listening rooms. Subjectively, these effects/limitations are easily audible as well. The way we arrange the speakers will interact with the inherent properties of our listening room in ways much more significant than much of the concerns "hardware audiophiles" often speak of or obsess over (like which CD player/DAC/streamer/server we use, cable differences, or if jitter even is audible :-).

The above should be obvious to readers already. Over the last month, with "social isolation" in place, there was time after work to try out something I've wanted to play with for awhile. Let's see if I can smooth out the low-frequency response in my room by experimenting with the use of an additional small subwoofer...

I. The issue with a single sub...

For reference, consider the previous work by Earl Geddes (nice summary here) and Welti & Devantier (2006). The idea of having multiple subs to handle low frequencies below 100Hz is well-discussed by now in 2020 and many home theater buffs and audiophiles already appreciate this. In summary, the hope is that by using multiple subs in a room, we will excite multiple room modes in a more "balanced" fashion so that in aggregate, the resulting frequency response will hopefully avoid deep nulls and very extended peaks.

For the discussion today, let's use my room as an example and see what could be done to improve things... Here's approximately what my room looks like as a floor plan:

For years, I've been running a single subwoofer - the excellent Paradigm Signature Reference SUB1 in the left front corner of the room. This sub IMO sounds excellent. Capable of clean, deep, unbloated sounds, power to literally shake objects off tables if pushed and the experience of the "compression" in the room can be to the point of discomfort. I almost never run it to such levels of course since my family would complain 2 floors up in the house if I turned this bad boy up at night in the basement :-).

At >100lbs and packing 1700W continuous power, I've always left the SUB1 in that location as a convenient spot given the other pieces of furniture and where my built-in cabinets are located in this room (mostly on the right wall). There's a problem with this location however which I haven't been happy with over the years. From the listening sweet spot on my sofa, there are sadly some rather deep nulls within the 40-70Hz region especially with signals in the left channel regardless of what I do to optimize the sub's internal DSP setting, adjust volume levels or twiddle with the variable phase control. The "best" I could achieve looks like this. (Note: No global DSP room correction applied, 1/12 octave smoothing for each side with 1/6 smoothing for average frequency response.)

Looking at the graph, we can easily see that the two speakers are reasonably congruent above 250Hz, but below that, the room plays a much bigger role in the sound. This transition starts around the "Schroeder frequency", usually for small rooms like this, somewhere from 200-300Hz. (As usual, I encourage you to measure your own room and see for yourself!)

With digital room correction applied (such as using Acourate as discussed over the years), I can smooth the peaks without too much effort, but those relatively deep nulls are another matter! Generally, it is not recommended to use strong EQ to boost these (typically) bass frequencies for various reasons - including the fact that adding unreasonable amounts of gain will cause strain to your amp and speakers, and potentially introducing distortions.

Let's consider then the idea of adding another sub to improve the bass response. I suspect that if one were to ask audiophiles about this, most might question the wisdom of doing so and suggest if one were to do such a thing, insist on getting another identical sub. While an identical sub might be ideal (not necessarily true unless proven so for the specific room!), it would both be rather expensive given the cost of the SUB1 but more importantly, this would be impractical as well! Where am I supposed to put another one of these in my room that would not get in the way of cabinets, doorways, or just for convenience?!

Thankfully, remember that the ears are not particularly sensitive to bass frequencies or directionality below around 100Hz (THX and home theaters typically set their crossovers at 80Hz). As such, "timbre matching" isn't really that important, and when crossed over low enough, we have good freedom of speaker placement.

So, let's try doing this with a much less expensive and smaller (slightly larger than 1' cubed!) option - the Polk Audio PSW111 which I already have at home being used for a small bookshelf system upstairs. Listed as US$300, you should be able to find these on sale for less. This is a small sub with only an 8" driver and reasonably powerful 150W continuous Class-D amplifier in the box. For its size, it digs quite deep with usable frequencies down to ~30Hz and I like that the sound is "fast" and not boomy when pushed up to moderate levels. It's a bottom-ported design and I believe the -3dB point is something like 42Hz. Remember, I'm not looking for another sub to enhance the lowest frequencies down to 20Hz since the Paradigm SUB1 can already do this, I just wanted a device that can interact primarily in the 40-70Hz range to reinforce the sound and hopefully in the process neutralize the deeper nulls there...

For completeness, as you can see in the pictures above, this little device can be fed with line-level RCA input as well as speaker level inputs (wire clips). I'll be using the RCA input (inexpensive shielded Amazon Basics subwoofer cable good enough so long as no audible noise pickup) fed from my Emotiva XSP-1 preamp while the main subwoofer (SUB1) is fed from the preamp's XLR output. Controls on the Polk include volume, variable low-pass filter adjustment between 60-120Hz, 0/180° phase switch, and the power switch can go into an "Auto" mode to save on electricity.

II. Hypothetically... Let's simulate the room for placement options first...

Okay, suppose that little sub might help, the question then is obviously, "Where should I put it to do the most good?". And for that, let's fire up Room EQ Wizard (I used version 5.20 beta 47/48 while working on this) and let's have a look at some "modeling" of the room.

Using "Room Sim", I can plug in the approximate size of my room, the characteristics of the speakers, and explore where I can put "Sub2" so that at least theoretically, the nulls will not be as deep.

There were a few locations I could find including the front right by the main speaker, and the rear left wall which looked like they would work based on the simulation. I ultimately settled on this placement based on the modeling results, listening with sub at the various locations, and for convenience:

Remember that Room Sim models a rectangular room whereas my space has a tapered front along with some irregularity of the right wall. As such, the simulation won't be exact, but we can see from the dotted "Ref" frequency curve that there are various deep nulls around 45Hz and 62Hz which correlates to some extent with the "Archimago's Room Response" left channel measurement above. If we were to add a second subwoofer with approximately the parameters of the Polk with LF -3dB around 40Hz located to the rear right of the room as circled, the predicted frequency response curve will change to the continuous black line. This looks beneficial in reigning in that ~30Hz peak while limiting the depths of the 45/62Hz nulls.

III. Placing the Polk PSW111 and measurements

Let's do it! Here's the little Polk sub ("Sub2") situated as per the simulation above beside the rear right surround Paradigm Studio 80 speaker; you can see the miniDSP UMIK-1 measurement microphone in the foreground. I've updated the room layout plan.

For listening and measurements, the electronic audio chain looks like this:

Roon running on Windows 2016 Server machine in another room --> ethernet --> Oppo UDP-205 --> XLR out --> Emotiva XSP-1 preamp --> Hypex NC252MP amplifier --> Paradigm Signature S8 v.3 front speakers 

Emotiva XSP-1 sub out* --> XLR --> Paradigm SUB1 

Emotiva XSP-1 sub out* --> RCA --> Polk PSW111 

* signals to SUB1 and PSW111 contain summed bass content

So how does it measure once I adjusted the subwoofers for relative volume, phase (Polk phase -180°, SUB1 0°), and crossover point (70Hz for Polk while SUB1 has internal DSP setting also around 70Hz)?

If we focus just in the frequency range where the little Polk sub would have most impact, we see that indeed between 35-90Hz, the major dips are "filled in". The left channel nulls at 47 & 68Hz have been neutralize. Likewise, the right channel peak at 78Hz has been removed (with only a small change in the 30Hz peak). Of course, it's not perfect but on the whole, the extremes have improved to a degree that will assist further optimization. These changes are already audible on tracks with smooth bass lines.

IV. Room correction DSP optimization

The next step is to create the Acourate room correction DSP filters. These are processed through Roon at playback (procedure as discussed here and here). In the last while, I've been playing with 4 filter "curve" settings as targets at my listening sweet spot which I can adjust on-the-fly depending on mood and the kind of music I'm listening to. Here they are:

I've got both "FLAT" targets as well as the psychoacoustically preferable "HARMAN"-like targets (see Toole, 2015). Note that for the "HARMAN" family, I do subjectively prefer a little more bass so what I did was introduce the taper from around 500Hz to 20kHz (I used a -7dB taper down to 24kHz), then added a boost from +1-3dB from 250Hz down.

The "DIP" room curves include a notch centered around 3.5kHz. As per this discussion with the late Siegfried Linkwitz, note how he discussed dropping the 3-4kHz region by 4dB to create a more "diffuse-field sound". This could help with many close-mic'ed recordings that could sound harsh in that "upper midrange" frequency region where our hearing is the most sensitive. For me, -3dB with some of my pop and 80's music sounded good especially with the overall "HARMAN"-like shape where the high frequencies are already gradually deemphasized. As I've discussed before, as a "more objective" audiophile, I think it's important to appreciate the stuff that can make a real difference through objective analysis, but also balance that by "knowing thyself" subjectively and optimizing the system to one's perceptions and preferences; nothing wrong with "euphonic" adjustments which is how I see the role of these "target curves".

DSP curves assigned as presets in Roon for easy selection/switching.

I won't bore you with the details around tuning Acourate settings since this will vary for everyone. I followed Mitch Barnett's recommendation when we chatted a little while back about using a smaller window for correction of the high frequencies to be beneficial. I also didn't want the correction filters to attenuate the signal too much so designed the room curves as above based on my measured response and allowed the filter generator to add up to +4dB gain where needed.

Here are the final pre- and post- room frequency response curves using the "HARMAN+DIP" curve as the target with dual subs as calculated by Acourate. For comparison, I also had a look at the effect of a similar filter using just the single Paradigm SUB1 without the little "helper" Polk sub:

Notice that having dual subs boosted the dips from 30-70Hz. This made it easier for the DSP to smooth out the frequencies <100Hz as can be seen if we compare the pink highlighted section. Note though that adding the little Polk sub did have side effects, such as a slightly bigger dip around 120Hz and the 25Hz dip is about 1dB deeper - nothing's perfect. ;-)

Another interesting observation we can appreciate here is that the Paradigm Signature S8 speakers appear to have incorporated a little "dip" around 3kHz already as highlighted in yellow similar to the -3dB 3.5kHz "DIP" purposely introduced in the target curve.

Remember that the above graphs are calculated "test" results created in the Acourate software. Let's finally remeasure the room again with REW and make sure that actual "real life" Roon playback with those filters applied resulted in the expected improvements. These measurements were made with the room as it would be when I listen to music - a small padded coffee table in front of the sofa, guitar in the corner and various other things like a stacked tower CD rack by my right front speaker.

1/6-octave smoothed average frequency response in green.

Looks good with reasonable correlation to the test convolution plotted above in the Acourate software.

To have a look at how sitting off-center changes the frequency response, here are measurements with and without DSP activated at the central sweet spot compared to 1.5' left and 1.5' right - did DSP treatment result in any more seat-to-seat variation?

1/6 octave smoothing for all graphs to make it not look "too busy".

As one expects, moving away from the sweet spot (in this case 1.5' to the right and left) will change the frequency response the person hears depending on the presence of those nulls and peaks at the location in the room. Whether DSP is engaged or not, notice that there are some strong peaks at 42 and 36Hz at 1.5' right and left of the central "sweet spot". On the whole, with DSP turned on, the graphs looked a bit less dramatic than without but I would say seat-to-seat consistency sounds about the same with or without DSP.

Let's finish off with a couple other measurements to check that there are no worrisome issues with the Polk sub in the system - let's start with the step response captured at the listening sweet spot (standard REW 500ms window):

Looking at -20ms before and +75ms after the initiation of the step, the time domain has been cleaned up significantly with the DSP active. Notice that the peak amplitude arrives by <5ms and the sound is "concentrated" within <15ms. In comparison, we can see that without the DSP, the energy is "smeared" with peaks approaching the strength of the initial deflection out to about 50ms. Remember again that this measurement is taken from the listening sweet spot about 11' from the main speakers with all the furniture in place; thus I'm not expecting the ideal "right triangle" step response, but rather evidence of improved time alignment (which appears to indeed be the case).

BTW, I think this is what actual time-domain "deblurring" looks like folks... Not whatever Bob Stuart and MQA were trying to sell us over the last number of years. ;-)

Regarding room reverberation, here's the RT60 with both right & left channels playing and feeding both subs, measured at the central "sweet spot" at the sofa again. The T60M (highlighted) is calculated from the detailed decay graph (not shown):

I would have liked the cyan EDT (Early Decay Time) to have been better below 200Hz. Along one of the walls in the room, I have a cabinet where I store some CDs, headphones, and other gear. The glass panel and cabinet door will rattle with deep bass and I suspect it's contributing to much of the low-frequency measured anomalies. Otherwise, the T60M is staying between 300-500ms from 100Hz up. Not bad I think for a domestic listening space although not ideal for music production control rooms (want to keep RT60 between 200-400ms for that purpose as per EBU3276 standard).

V. Listening...

Contrary to certain "subjectivists" who seem to think objective-leaning folks don't listen (!!!), I did of course make sure to listen to the differences adding the little Polk sub made. In fact, this project took a number of evenings to fine tune and the write-up above is simply a summary sparing the reader the trial-and-error details. Using data from the room simulations, not only did I listen but also made quick measurements at 3 different sub placement positions before settling on the final placement above. It was only once I settled on the final placement, did I fine tune the phase and adjust relative volume levels between the Paradigm SUB1 and Polk PSW111. From there, I did the Acourate room sweep recordings and constructed the various DSP filters based on personal listening preferences developed over the years... Finally, confirmatory REW measurements including the time-domain step response and off-center measurements to complete the investigations.

As I've expressed before, it's not hard for a pure "subjectivist" to write and review products, or claim that they "hear" differences and details. Admittedly, it's a lot more effort to drag out the measurement microphone and perform the objective "work" but IMO it's worth it both intellectually and learning to appreciate that the objective and subjective elements of this hobby go hand-in-hand. There's certainly a "joy" in performing this exercise as a "hardware audiophile". Ultimately, to optimize fidelity, one will need to do one's own homework since nobody else will have your room, your gear, and your subjective preferences.

As I finish writing this on my laptop while listening to the final result, I like what I'm hearing. The beautiful percussion work on a classic audiophile track like "Drum Improvisation 1" off The Sheffield Lab: Drum & Track Disc is presented with a nice wide sound stage across the front with instruments of the drum kit nicely laid out and projecting a subjective sense of depth. With so much of our music synthetic or heavily processed these days, it's a joy to listen once awhile to naturally recorded material such as this to remember the experience of high dynamic range and the feeling of the visceral "attack" of a bass drum played at reference volumes. Another naturally recorded album with tons of bass would be Reference Recording's Organ Sensation Felix Hell. Of course, almost any large organ recording will likely have prodigious bass when reproduced well.

One of my favorite tracks from UB40 is "Kingston Town (Extended Mix)" which I got as a single back when Labour Of Love II was first released. Awesome sounding bass line which is easily missed if one's system has limited low-end reproduction. If one is more interested in electronic music, Melleefresh vs. Deadmau5's "Whispers" (Cocktail Queen) sounds great with a nice system... Oh yeah, of course Yello's stuff like "Planet Dada (Flamboyant)" (The Eye) is a nice demo track as well.

Finally, thanks to Mitch Barnett for this article that dissects the bass fundamentals in Rebecca Pidgeon's "Spanish Harlem" (The Raven) which we can use to listen for unevenness below 100Hz.

VI. Conclusions

I like what I hear with the dual sub set-up so far and probably will listen to this for the next while before trying anything else to further optimize the system ;-).

Based on this experiment, I do believe excellent results can be obtained with relatively inexpensive tweaks like the addition of this little sub. I don't hear or see anything wrong with mixing and matching different sized subs or brands so long as we can appreciate the limitations of the device in achieving what we want and carefully tuning the effect - ideally guided with objective measurements. For example, with the little Polk PSW111, I'm not going to be pushing its volume to the max (nor do I need to even at reference levels). I know this sub will not reach below 30Hz which is why it's being paired with the much more power Paradigm SUB1 for those deep deep rumbles. I'm also keeping the low-pass crossover set around 70Hz to target the effect within 30-70Hz and reduce the chance of speaker localization.

Ideally, it would be nice to have better control of the subwoofer crossovers and even digital correction filters for the subs rather than a global filter like this. Obviously that would take extra work and complexity which to be honest might or might not be worth the time and effort given the improvement I'm hearing.

Speaking of localization effects, although I cannot prove it, I did find it preferable to place the little Polk to the rear right while the larger SUB1 stayed on the left front. One of my experiments was to place the Polk at the far left rear corner of the room. While frequency response measurements looked fine, subjectively there just seemed to be an excessive left-sided "heaviness" to the sound in my listening tests. Like I said, I can't prove this since it could all be psychological knowing that both subs were located by the left wall.

Have fun experimenting!

For those who have never taken a measurement mic to your room (miniDSP UMIK-1 highly recommended BTW), I consider it a bit of an "audiophile rite of passage" to do this as only then can one truly, and personally appreciate the effect of your room. Once we appreciate the dB variations in the frequency response, the time domain irregularities, and perhaps how reverberation time might deviate from the ideal in our rooms, then I think we can also truly contextualize just how little difference some of our components (like cables perhaps!?) actually contribute to the final sound!

--------------------

OMG - this is ridiculous. I see that about a month ago, John Darko took another stab at claiming significant audible differences between digital cables with this video. As I mentioned a few weeks back, I do think there is a big difference between the "experts" in the audiophile niche he's polling and those engineers who might have worked on developing/implementing the USB standard (from the likes of Microsoft, Intel, AMD, Apple, XMOS, TI...). In court for example, experts disagree on all manners of things all the time! The fact that he's only advancing the opinion of one set of partisan "experts" isn't really all that helpful in the absence of any kind of evidence! It's rather telling that he turned off the comments section on that YouTube video... Gee, I wonder why?

Seriously folk, unless we see some evidence for things like USB cables making a difference, an appeal based on the opinion of "experts" without reference to any documented science is simply worthless. There is no "period pain" (apart from our current "pandemic period pain" I suppose). There is no change in paradigm coming. "Flat Earth" and "Bits Are Bits" are not analogous whatsoever. Proprioception and the idea of a spatial body schema are also not new concepts BTW that popped up since he left school.

Based on my private conversations with a few respected audio industry engineers over the years, I do not think many actually believe boutique USB cables are at all necessary even if they don't say this publicly. Nobody has presented any evidence that USB cables significantly change sound quality from a DAC. If one needs an analogy, rather than using the idea of a "Flat Earth", I would submit that those like Darko who vehemently support USB cables making a significant difference are more like "UFO Believers". Yes, they provide all kinds of testimonials. Sometimes they even provide "footage" (like the blurry Pentagon UFO videos lately) of demos where something sounds different, but they don't examine the "evidence" deeper to look for shenanigans like changes in volume. Like the belief in UFOs, it's "fun" and for the most part "harmless" of course so long as one doesn't go overboard and get shot trespassing into Area 51 or in the case of audiophile cables, one doesn't mind paying the "snake oil premium" over otherwise good generic products.

Darko's testimonial video is simply an example of wishful thinking, psychological susceptibility, and a desire to promote products consistent with the wishes of his advertisers and the Industry IMO. By all means, try boutique USB/digital cables for yourself and do it blindly if you can within a return period if the product is expensive. I think very quickly you'll realize that all this stuff he's talking about can be (and should be) safely ignored like UFO or other paranormal testimonials you might have encountered over the years.

[BTW, as a sci-fi fan, I would love to witness humanity making "first contact" with extra-terrestrial life! While obviously much less dramatic, I would likewise love to see evidence that this US$800 1.5m AudioQuest Diamond performs better than this US$6 1.8m AmazonBasics USB "printer" cable connected to a high quality DAC...]

In closing, I want to thank everyone who took part in the latest "Is high harmonic distortion in music audible?" Internet Blind Test. The number of responses has reached >60 participants thus far and while I said I was going to close it off at the end of April, just in case there are some stragglers, I'll close off the survey after this weekend so get any last minute submissions in!

Have a happy and safe May, everyone. I need to get some server hardware upgrades and backups done this coming week - updating to Windows Server 2019 as part of the change and installing a very good (so far) yet inexpensive TP-Link AX3000/AX5 WiFi 6 router. With society slowly opening up post-first-wave COVID-19 around here, there will also be some planning and adjustments at work I need to address so there might not be as many blog posts this month although I will of course analyze and summarize the results on that Internet Blind Test :-).

Hope you're all enjoying the music!